Smart Solutions for Insulator Condition Monitoring by Hector de Santos
Severe environmental and industrial pollution is a concern for electric utilities due possible resulting flashovers and unplanned line outages. In this regard, insulator condition monitoring is a valuable tool to allow maintenance actions, such as washing, to be scheduled when needed. Among the variables that can be monitored, leakage current stands out as the most meaningful since it provides a true measure of how close an insulator string is to flashover. The relationship between leakage current and environmental as well as climatic factors that can impact insulators has therefore attracted much attention. But since this relationship is complex and dynamic it cannot be successfully depicted using mathematical tools. This presentation proposes a new solution for condition monitoring of insulators based on estimating leakage current from environmental and meteorological data with the help of machine learning. Such methods are a branch of Artificial Intelligence and allow computers to ‘learn’ from data. Set-up of different condition indicators is supported by leakage current data obtained from artificial pollution tests in a laboratory. Condition indicators can be also configured to measure Site Pollution Severity or to assess ageing of RTV silicone coatings. Results from a 3-year monitoring project in an outdoor test station show high accuracy when comparing estimated and actual condition indicators.